Exercise influences different endocrine and metabolic parameters, and information in the literature is sparse for some of these hormones and metabolites in the exercising horse. The aim of the present experiment was to study the metabolic response to exercise when feeding diets with varying carbohydrate composition (fibre and starch) under experimental conditions where diet and exercise were standardised and controlled. The response was investigated in a 4×4 Latin square design experiment using four Norwegian Coldblooded trotter horses. The dietary treatments were two fibre based diets, hay only and hay (85% of dry matter intake (DMI)) supplemented with molassed sugar beet pulp (mSBP) (15% of DMI), or two starch based diets of hay (68% of DMI) and barley (32% of DMI), and hay (68% of DMI), barley (26% of DMI) and mSBP (6% of DMI). Each diet was fed for 28 days; 16 days of adaptation followed by 12 days of data collection. 4 h after the morning feeding at 06:00 the horses performed a standardised exercise test (SET) lasting 45 min. Blood samples were taken before feeding the morning meal at 06:00, before the SET (10:00), after the SET (10:45) and after recovery from exercise (15:00), and plasma samples were analysed for relevant metabolites and hormones. Plasma leptin concentrations increased after exercise but were not affected by diet, whereas diet and exercise had no effect on the plasma concentrations of ghrelin and insulin like growth factor 1 (IGF-1). Furthermore, diet influenced the plasma concentrations of short-chained fatty acids (SCFA) more than exercise. The results provide important comparative information that can be useful in studies where diet and exercise cannot be controlled, e.g. in field studies.